On the integrability of the sine-Gordon system
This thesis investigates the integrability of the sine-Gordon system of nonlinear partial differential equations when the dependent variables are subject to some very particular boundary conditions. In chapter 1 the sine-Gordon system is introduced and, with N ϵ Z, P, Q ϵ R, the sets of initial-boundary value problems A(_N) and B(_P,Q) are defined. In the set A(_N) at the spatial variable x is unbounded and the boundary conditions are fixed by initially choosing the topological charge N. This set of problems is the one usually associated with the sine-Gordon system. In the set B(_P,Q) the spatial coordinate is constrained to the semi-line (-oo,0) and there exists two boundary parameters P,Q ϵ R to be chosen a priori. It is the study of this second set of initial-boundary value problems for arbitrary P, Q which forms all the original work of this dissertation. The study presented here is primarily concerned with the development of three separate inverse scattering methods for solving these sets of initial-boundary value problems. The first of these is developed in chapter 3 and is applicable to a subset of the problems in A(_N). The method is the one usually associated with the sine-Gordon system and studies the asymptotics of the initial data as x → ±oo. It is included in this thesis for completeness and as background for the original material which follows. Next, in chapters 4 and 5, the inverse scattering methods appropriate to initial-boundary value problems in subsets of B(_P,O) and B(_P,Q#O) are constructed. In these cases it is important to realise that it is only possible to study the asymptotics of the initial data as x → -oo. Once these three methods have been formulated they are used to find soliton solutions and infinite sets of integrals of motion for these boundary value problems. When a boundary is present at x = 0 the interaction of the solitons with this boundary is studied. These topics are addressed in chapter 6. Finally in chapter 7 the question of the integrability of both sets of problems is addressed. By interpreting the various inverse scattering methods in terms of canonical coordinate transformations of phase space it is seen that the existence of such methods can be viewed as a constructive proof of the integrability of these boundary value problems.